Comment 4 for bug 167076

Revision history for this message
Richard Hughes (cyreve) wrote :

I've been looking at this bug for the past few days without
doing anything about it, but now Bulia's assigned it to me I
suppose I've got to admit that I haven't got a clue what the
problem might be. The following text, therefore, is composed
of 50% random guesswork and 50% grasping at straws.

- Try wiping your font caches and starting again:
  cd /usr/share/fonts
  find . -name fonts.cache-1|xargs rm -f
  fc-cache .

- I have a font called "Lucida Bright" which works for me on
0.42.2 (Win32), however I think it looks slightly different
to yours. lbrite.ttf 70,748 bytes, md5
825a2395154f2a944b653bcb7839dd27

- The only part of the code which I think is unusual is that
we force all fonts to be 'outline' (whatever that means: I
don't know much about the internal structure of ttf either).
If a font is not outline then we'll sometimes fail to load
it and sometimes coerce the outline property on, possibly
causing a different font to be selected.

- Your Chicago screenshot looks particularly interesting
because of the massive amount of kerning on the T's. Since
the real Chicago has quite narrow T's, this could mean that
the correct metrics are being used but the wrong glyphs. Can
you try some other characters to see if you can figure out
if that is the case, or if it's just normal kerning?

- I might think of something else to add later. We'll see.